We estimated a Non-Stationary Dynamic Factor model and used it to generate artificial epi-sodes of disinflation (permanent change in the mean inflation rate). These datasets were used to test the forecasting abilities of alternative underlying inflation indicators (i.e. the measures that capture sustained movements in inflation extracted from information in a disaggregated set of price data). We found that the out of sample forecast errors of the bench-mark underlying inflation measures (based on unobserved trend extraction) are more severely affected by disinflation than the alternative simpler methods (based on exclusion or reweighting approaches). We also show that a Non-Stationary Dynamic Factor model may be employed for extraction of the unobserved trend to be used as an underlying inflation measure.